Analysis code and results for: Empirical validation of the Consciousness Potential framework
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Companion code for:Howie, S. J. (2026d). Empirical validation of the Consciousness Potential framework: Developmental trajectories in cultured cortical networks and clinical group differences in dementia. Zenodo. https://doi.org/10.5281/zenodo.19362803 OVERVIEW This deposit is the analysis code and canonical result tables for the empirical-validation paper. It covers three lines of evidence, each reproducible from bundled data: MEA developmental trajectory (Section II): Phi and its four components computed across days-in-vitro on Wagenaar dense-culture cortical-network recordings; produces the developmental-trajectory figure. Single-cohort EEG (Sections III.1 to III.4, OpenNeuro ds004504): per-subject Phi and C x 100, HC-vs-AD and HC-vs-FTD group differences, the within-AD MMSE correlation, and a region-specific frontal-to-temporal Granger analysis. Multi-cohort replication (Section III.5): a three-cohort directional replication on ds004504 + BrainLat + figshare 5450293 (n = 266). Per-component cross-cohort effect sizes, HC-anchored rank-normalization verification, within-AD severity stratification, and FTD-vs-AD direction comparison. The honest composite-scalar result is near-null (pooled |d| about 0.1); the surviving signal is a two-component direction (I_A down, organization up; pooled |d| about 0.43). The recursive feedback index I_R is the normalized spectral radius of the thresholded Granger-causality matrix, the framework's sub-critical estimator for the returning mutual information specification of I_R. The reachability-ratio proxy is retained only for traceability. SELF-CONTAINED REPRODUCTION bash run_paper4_v7.sh This runs multicohort_canonical_section35_v1.py first, which regenerates the Section III.5 canonical numbers directly from the bundled cohort caches outputs/ds004504_5d_car_data.json and outputs/brainlat_5d_car_data.json, with no external deposit required. It prints VERIFY OK when the reproduced figures match the manuscript. The Section III.3 single-cohort statistics reproduce from the bundled eeg_phi_results_calibrated.csv via eeg_analysis_final_v2.py, and all Section III.5 headline numbers are also inspectable in the bundled paper4_v7_*.csv tables. The earlier ep5 multicohort scripts are retained as legacy paths that recompute from the paper-7 deposit feature JSONs; they run only when a --data-root or PAPER4_DATA_ROOT is supplied. CONTENTS (selected) multicohort_canonical_section35_v1.py — Primary, self-contained Section III.5 from the bundled outputs/ caches.multicohort_ad_replication_v1.py, multicohort_component_analysis_v1.py, multicohort_signature_verification_v1.py, multicohort_ranknorm_verification_v1.py — Legacy Section III.5 scripts (need the external paper-7 feature JSONs).eeg_phi_analysis_v3.py, eeg_analysis_final_v2.py — Single-cohort ds004504 Phi plus statistics.eeg_regional_analysis_v2.py — Region-specific I_M and Granger (Section III.4). LEGACY pipeline: it computes the retired I_M-product Phi and reports its C x 100 on the legacy raw-Phi scale (local PHI_REF = 0.015647); the paper cites only its regional I_M and Granger outputs, not that C x 100.phi_approx_mea_v2.py — MEA developmental trajectory (Section II).framework_core/ — Shared constants, signal IO, and Phi-component computation (post-retirement four-channel Phi = I_A x (1 - H_norm) x I_R x I_P).run_paper4_v7.sh — Pipeline driver.paper4_v7*.csv, eeg_phi_results_calibrated.csv, eeg_regional_results.csv, phi_approx_mea_per_div.csv, phi_approx_developmental_trajectory.png — Bundled canonical result tables and the figure. REQUIREMENTS Python 3.8 or newer with numpy and pandas. EEG file reading and statistics are hand-rolled from NumPy primitives, so no scipy or MNE dependency is needed for any cited-reproducible path. HEADLINE NUMBERS Cohorts: n = 266 (HC 29 + 32 + 100; AD 36 + 35 + 34) across ds004504, BrainLat, figshare 5450293.Composite C x 100 HC-vs-AD: pooled |d| about 0.1 (honest near-null at the scalar level).Two-component direction: I_A down, organization up; pooled |d| about 0.43.Within-AD MMSE Spearman rho = -0.43. CROSS-DEPOSIT NOTE This deposit's _framework_core implements the post-retirement four-channel canonical Phi = I_A x (1 - H_norm) x I_R x I_P (persistence I_M retired). The geomean anchor PHI_REF = 0.716 is flagged stale pending a series-wide recalibration. Raw Phi and C x 100 values are not directly comparable across deposits that ship different framework-core vintages (the Paper 5 deposit ships a five-channel I_M form). VERSION v5 (June 2026): adds the self-contained multicohort_canonical_section35_v1.py and the bundled outputs/ cohort caches so Section III.5 reproduces from deposit-only state; flags eeg_regional_analysis_v2.py as legacy with a corrected local PHI_REF; documents the Phi vintage. Tracks the paper v7 numbers, unchanged through paper v8. Version 3 (May 2026) added the multi-cohort analysis, the spectral-radius I_R, and the consolidated _framework_core module. CITATION Software (version-independent concept DOI): 10.5281/zenodo.19362255Companion paper (concept DOI): 10.5281/zenodo.19362803
Fields of Study & Subjects
Neural dynamics and brain function
Functional Brain Connectivity Studies
Philosophy and Theoretical Science
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 1 April 2026
- Language
- EN